RFM Analysis

Zoning and Cloning your Best Customers; Avoiding the Worst

Before the bubble popped in April of 2000, all that mattered for a new economy company was to get big fast, which in turn meant acquiring customers at any cost.

Well, Toto, we aren’t in Kansas anymore— which is my way of saying that start ups must now prove they are on a “path to profitability”. This in turn is driving people to go back to the future, to return to fundamental tools of direct marketing, such as RFM. RFM stands for recency, frequency, and monetary value and is one powerful yet relatively simple analytic tool.

RECENCY (R)
This is a score designed to represent how recently the customer bought from you. All things being equal, customers who purchased from you most recently are the best source of new business for three reasons:

Sanitary data
The more recent the buyer the better the chance that the data corresponding to that buyer is update and in good hygene, which means you’ll get fewer pieces of email or snail mail returned to you. Email addresses in particular age very quickly, which means that if you depend on email to drive repeat business, less recent names on your file are likely to yield far less revenue that more recent names.

Brand affinity
Recent buyers were attracted to your brand for a reason, and presumably that reason will extend to another purchase that is close in time to the first.

Actively seeking solutions in your category
A recent buyer generally corresponds to a buyer who is active seeking solutions in your category. (If you’ve changed strategic directions and don’t feel recently is relevant to active search behavior in the category, you may wish to go on and look at other factors, such as product affinities score. See David Shepard’s book for a discussion of product affinities and how to calculate them.)

Scoring
One way to score a file for recency is like so:

Give all customers who have purchased in the last 6 months a score of 3
Give customers who have purchased within 6 to 18 months a score of 2
Give customers who purchased within 18 to 36 months a score of 1
Give all customers on your file for 36 months or more get a score of zero (which is about what they’re probably worth)

FREQUENCY(F)
The frequency that a given customer purchases from you in a particular time period. Often times, this means recording how often a particular customer purchases within a 90-day window. Some people also like to score the file based not only on frequency in a given time period but also count the number of categories a particular customer has purchased in during that same window. This gives you a measure of depth of purchasing activity within your overall franchise. The idea being that if you have a large and broad product base, both frequency and depth will be important in separating out the best customers from the rest.

MONETARY VALUE (M)
The value of that customer’s purchases, ideally in terms of net margin or profit contribution.

Continuing with our scoring example
Now multiply all the scores together (R x F x M). Sort the file with high RFM customers at the top and low RFM customers at the bottom. The top 20 percent of your file represents your zone of opportunity. These are your best customers. The bottom 20 percent of your file represents your zone of avoidance. These are your worst customers.

This new information lets you do two important things. First, you can prioritize your marketing resources based on objectives for each of these customer segments. Many successful companies spend heaviest against the middle 60 percent, focusing on moving them up to the top 20 percent. Second, you create a profile of your best customers and use that information to clone these types of customers next time you field a marketing program designed to acquire new customers.

Cloning Customers
Say you have a file of 1 million records. The top 20 percent of customers based on RFM represents 200,000 records. An old homily from direct marketing is that the best source of new business is customers who look like your old customers. Consequently, what you want to do is take those 400,000 records from the top and bottom of your file and match them to one of several published database sources, to learn more about them. For example, if you sell business-to-business, you may wish to match the file to Dun & Bradstreet’s file, to be able to profile your best customers by company size and SIC (standard industry classification) code. If you sell to the home market, you may want to work with Metromail or Lifestyle Selector, who have rich databases that can give you both demographic and psychographic information on your customers.

Now that you have identified your best customers and what they look like in terms of the types of businesses they work in or the demographics and psychographics that best describe them, you start to target your marketing mix toward acquiring more customers who fit the profile of what your best customers look like.

Avoidance as a strategy
What about your worst customers, those customers at the bottom of your file with the lowest RFM scores? One of the realities of life after the bubble popped is that companies can no longer afford to waste dollars chasing after unprofitable customers. Which means you can use the profiles you’ve developed that describe the bottom 20% of the file – to ensure you don’t go after these types of customers in your next marketing campaign.

First published in Marketing Computers magazine in September 1995 and updated July 2001; reprinted here with permission.